ASSESSMENT OF VENOUS THROMBOSIS IN ANIMAL MODELS

Abstract

Deep vein thrombosis and common complications, including pulmonary embolism and post thrombotic syndrome, represent a major source of morbidity and mortality worldwide. Experimental models of venous thrombosis have provided considerable insight into the cellular and molecular mechanisms that regulate thrombus formation and subsequent resolution. Here we critically appraise the ex vivo and in vivo techniques used to assess venous thrombosis in these models. Particular attention is paid to imaging modalities, including magnetic resonance imaging, micro computed tomography and high frequency ultrasound that facilitate longitudinal assessment of thrombus size and composition.

INTRODUCTION

Deep vein thrombosis (DVT) is a common condition with an annual incidence of approximately 1 in 1000 in the general population and can lead to fatal pulmonary embolism (PE). Together these conditions account for a greater number of deaths in the United Kingdom than those caused by breast cancer, road traffic accidents, and AIDS combined ^{1, 2}. Approximately one third of patients with DVT develop post-thrombotic syndrome, a chronic condition characterised by persistent limb pain, swelling, and ulceration, which carries a significant health and economic burden, and is associated with a reduced quality of life ³.

Treatment of DVT currently involves systemic anticoagulation, aimed at preventing secondary thrombotic events, and catheter directed thrombolysis, which in the acute setting has proven effective in removal of the thrombus. However, both of these therapeutic strategies give rise to pathological bleeding in a significant number of cases and may be contraindicated in specific patient subpopulations. Novel treatments that either prevent thrombus formation or hasten resolution without these side effects are desirable, and are likely to arise from a better understanding of the molecular and cellular mechanisms that control venous thrombosis.

Experimental models of DVT have been developed in a variety of animals, including the mouse ⁴, rat ⁵, rabbit ⁶, dog ⁷, pig ⁸, and non-human primates ⁹. Spontaneous, symptomatic DVT is not observed in these animals with researchers instead relying on a number of physical or chemical interventions on a given vessel (such as ligation or ferric chloride) to induce thrombosis. Although the coagulation and fibrinolytic systems in non-human primates most closely resemble those in man, the use of these species present researchers with both financial and ethical dilemmas. Other large species such as pigs have a similar fibrinolytic responses to man, lending themselves to assessment of thrombolytics whilst the close resemblance of the coagulation system in sheep to that of man may be of particular utility when assessing novel antithrombotic agents ¹⁰. However, murine models particularly those involving the infrarenal vena cava (IVC) currently predominate; owing to their technical simplicity, compatibility with available imaging platforms and the availability of transgenic strains in this species. These murine models have proven especially useful in elucidating the molecular and cellular determinants of venous thrombosis ¹¹.

Comparison of data from different studies is complicated by the variety of models used (both animal species and mechanisms of induction) in conjunction with the wide array of analytical techniques employed. Whilst the relative merits of respective models have been the source of critical review ¹² there remains no consensus on the best method(s) to accurately quantify venous thrombosis in the pre-clinical setting. In this review we critically evaluate ex vivo and in vivo methods currently used in the assessment of venous thrombosis and describe emerging imaging techniques that may prove useful in studying this dynamic condition.

EX VIVO ASSESSMENT OF VENOUS THROMBOSIS

Weight

Thrombus weight is a simple, quantitative and inexpensive measure of thrombus formation and subsequent resolution; with weight decreasing as the thrombus resolves (Figure 1A). Thrombus weight can be obtained with or without excision from the surrounding vessel ^{13, 14}. Measurements of thrombus weight without excision from the vessel may introduce variability associated with the inclusion of the vein wall and extraneous adherent tissue. Excision of the thrombus provides a more direct measure of weight, but during later stages of thrombus resolution, thrombus and vein wall become difficult to separate. Importantly, measurement of thrombus weight permits further biochemical and cellular analysis of the thrombus. It has been proposed that adjusting for thrombus length reduces apparent intra-group variability when measuring thrombus weight ^{15, 16}. In our experience such adjustments are not always sensitive to changes in thrombus size; for example, when thrombus weight decreases proportionally to thrombus length, similar values of length-to-weight ratio will be obtained.

Figure 1. Weight and morphology of resolving murine venous thrombi.

(A) Thrombus weight in the St Thomas’ model of IVC stenosis measured at days 1, 4, 7, 10 and 14 post- induction, bars represent mean ± SEM. (B) Representative micrographs of transverse thrombus sections stained with Martius scarlet blue (MSB) at days 1, 7, 14, 21 and 28 post-induction. MSB detects collagen (blue), fibrin (red) and erythrocytes (yellow), scale bars 200μm (low power) and 25μm (high power). Adapted from Saha et. al. 2013 ²³.

Histological analysis

Histological techniques are commonly used to provide estimates of both thrombus size and composition. Thrombus is excised in situ with surrounding vein wall and prepared for histological sectioning. Estimation of thrombus size is carried out by analysis of thrombus cross-sectional area in sections taken at set intervals (300-500μm) along the entire length of the thrombus. The summed area multiplied by the distance between sections provides an estimate of thrombus volume ^{17, 18}. We believe that reconstitution of thrombus volume provides the best estimate of overall thrombus burden. Measurements of representative cross-sectional area or scoring for the presence of thrombus at set intervals may not adequately account for changes in thrombus length or area respectively ^{4, 19, 20}. Histological analysis does, however, present a number of technical challenges. Sectioning of thrombus, particularly at early time-points, can prove difficult because of the friable nature of this tissue. Sectioning individual thrombi at defined intervals along their length, with subsequent staining and analysis of sections at each level is also time consuming. A further limitation of this technique is that processing of thrombus for wax embedding results in significant shrinkage that may distort differences between groups ²¹.

Histological preparation of venous thrombi also enables parallel analysis of thrombus composition by either tinctorial or immunohistochemical staining. Thrombus resolution is characterised by extensive remodelling of the extracellular matrix by cells that infiltrate the thrombus. Extracellular matrix deposition can be readily detected with a number of tinctoral stains such as Picrosirius Red (collagen), Van Giessen (elastin), Alcian Blue (proteoglycans) and Martius Scarlet Blue (fibrin and collagen, Figure 1B). Although these techniques are well established, conditions must be tightly controlled to ensure tissue structures are appropriately and consistently stained. Immunohistochemical localisation of the cellular infiltrate (e.g. leukocytes, endothelial cells, myofibroblasts) can also be used to assess thrombus organisation (Table 1). Careful consideration of cell specific markers and extensive optimisation of antigen binding are required.

Table 1. Cell types present in the venous thrombus.

In murine models of vena cava thrombosis, ‘early’, and ‘mid-late’ stages of thrombus resolution refer to days 1-7 and days 7-28 post-induction respectively.

Cell type	Marker (antibody clone)	Infiltration	Reference
Platelet	CD41 (MWReg30)	Early	37
Neutrophil	Ly6G (NIMP-R14 or 1A8)	Early	18, 37
Macrophage	Mac2 (M3/38), F4/80 (CI:A3-1 or BM8)	Mid-late	16, 17, 37
Endothelial	vWF (F8/86), CD31 (JC70A), Isolectin-B4 (GSL I-B4)	Mid-late	17, 65, 66

IN VIVO ASSESSMENT OF VENOUS THROMBOSIS

The terminal nature of ex vivo assessment is a significant limitation in the study of thrombus resolution as it precludes longitudinal measurements of changes in thrombus size in the same animal. Imaging techniques that facilitate longitudinal and reproducible quantification of thrombus burden overcome this limitation and provide powerful analysis through the generation of paired data. This approach can also limit the number of animals required for experimentation consistent with the principles laid out by the national centre for the replacement, reduction and refinement of animals in research.

Magnetic resonance imaging (MRI)

Non-contrast MRI

MRI can be used to diagnose DVT in man without the need for contrast agents ²². MRI time-of-flight venography is a rapid contrast-free technique for assessing thrombus volume in pre-clinical studies²³. Phased sequences are used to allow specific visualisation of the venous system, where thrombus appears as a flow deficit that can be segmented to allow quantification of thrombus volume (Figure 2A). MR longitudinal relaxation time (T1) mapping can also be used to image thrombus, which has a shorter T1 compared with surrounding blood and tissue ^{23, 24}. Shortened T1 times are likely caused by the accumulation of paramagnetic iron (Fe³⁺) in the thrombus²³. Temporal changes in thrombus T1 have been observed during murine venous thrombus resolution and are associated with increasing thrombus organisation (Figure 2B) ^{23, 25}. Magnetization transfer and diffusion weighted MR sequences have also been developed to characterize the age and protein composition of the thrombus ²⁶. Combining the T1, magnetisation transfer and diffusion weighted sequences, whilst time consuming to acquire, could provide valuable information of both thrombus size and organisation. A major benefit of non-contrast MRI is that newly developed sequences are rapidly translatable from the laboratory to the clinic without need for regulatory approval. The time restrictions imposed by MRI, and multisequence imaging in particular, limit the utility of this technique to assessment of thrombus resolution that occurs over a period of weeks, rather than formation in which the thrombus forms in the order of minutes.

Figure 2. Imaging of the resolving venous thrombus by MRI.

(A) Venous phase time-of-flight scans used for imaging of the IVC, presence of thrombus results in a filling defect in the vessel which can be used to estimate thrombus volume. (B) Generation of T1 maps demonstrates a temporal shortening in T1 relaxivity (red shift) as the thrombus resolves. Adapted from Saha et. al. 2013 ²³.

Contrast MRI

Agents, such as gadolinium, have been used to image thrombi in experimental models. In a baboon model of IVC thrombosis, gadolinium enhanced MR venography was effective in identifying the thrombosed segment ⁹. A number of peptide-conjugated contrast agents have been developed that target matrix components present in the acute and resolving venous thrombus, such as fibrin and collagen ^{27, 28}. A fibrin targeted contrast agent, EP-2104R, has been used to visualise intracranial thrombosis in pre-clinical models of stroke ²⁹ and is currently under evaluation for the detection of thrombi in man ³⁰. Furthermore, EP-2104R allows visualisation of the component of the venous thrombus that is susceptible to lysis in a low flow mouse model and could find clinical utility in stratifying patients for thrombolysis ³¹. A significant limitation of contrast-based approaches is the need for extensive validation of target specificity and the need for regulatory approval before use in human subjects, requiring costly clinical trials. Use of MRI in the pre-clinical setting has a number of limitations that include; the availability of scan time on much in demand clinical 3-Tesla scanners, the cost of scan time and the extended length of time required for an individual scan.

Micro computed tomography (microCT)

Technological advances in the field of computed tomography (CT) have facilitated the development of high-resolution microCT imaging platforms suitable for pre-clinical use. Contrast-enhanced microCT has been used extensively in the study of murine models of cardiovascular pathologies including critical limb ischaemia, abdominal aortic aneurysms and myocardial infarction ^32-34.

Visualisation of the vasculature by microCT requires intravenous administration of high molecular-weight blood-pool contrast agents such as Iopremol, ExiTron nano 12000 or Aurovist ³⁵. Contrast enhanced microCT enables longitudinal measurements of thrombus resolution as demonstrated in a murine model of IVC stenosis ³⁶. This technique requires segmentation of thrombus from surrounding tissue, allowing 3D reconstruction and extraction of volumetric data. A strength of contrast-enhanced microCT is the ability to obtain high-resolution images with reconstructed voxel dimensions in the range of 20-60μm ³⁶. Limitations of this technique for thrombus imaging include the high doses of ionising radiation required, the use of expensive and potentially nephrotoxic contrast agents and the cost of the imaging time. In addition, contrast enhanced microCT provides only anatomical data and not information on thrombus composition as can be obtained by MRI.

The use of fusion imaging modalities, such as CT-fluorescence resonance tomography (CT-FMT) enables the concurrent collection of anatomical and biological data in a non-invasive manner. CT-FMT provides imaging of cells, proteins and enzymatic activity through the use of targeted and activatable near infrared (NIR) probes. This technique has been used to evaluate thrombus composition with respect to macrophage and fibrin content, and matrix metalloproteinase activity localised to the thrombus ^{38, 39}. Angiosense a NIR blood pool contrast agent used for quantification of tumour vasculature may also be relevant to studies of venous thrombus neovascularisation ⁴⁰. FMT allows for fast acquisition of data with scan times in the range of 5-8minutes but affords limited spatial resolution in the sub-millimetre to millimetre range dependent on object depth ^{38, 41}.

Radionucleotide Imaging

A number of targeted radionucleotide probes have been developed to localise components of the thrombus in vivo. Initial efforts to visualise fibrin utilised labelled components of the fibrinolytic system including ^99mTc-tissue type plasminogen activator, ⁶⁷Ga-urokinase type plasminogen activator and ^99mTc-fibrin fragment E1, imaged by single photon emission computed tomography (SPECT) in a rabbit model of venous thrombosis ^{6, 45, 46}. Integrin αIIbβ3 has also proven to be an attractive target for the generation of SPECT probes with labelled disintegrin-like peptides localising to thrombus in a canine femoral vein model ^{7, 47}. Alternatively, positron emission tomography CT (PET-CT) can also be used to image venous thrombi. The PET tracer ¹⁸F-fluorodeoxyglucose localises to the acute thrombus in a novel model of recurrent DVT ⁴². Other targeted PET probes developed for imaging of arterial thrombi and atherosclerosis, which bind fibrin and the platelet surface receptor glycoprotein VI, may be directly applicable to the study of venous thrombosis ^{43, 44}. Advantages of SPECT over PET include; simpler probe generation, longer prober half-lives and higher-resolution images (approximate voxel dimensions of 350μm and 850μm respectively). However, PET provides higher sensitivity compared to SPECT that may be of particular importance when quantifying fibrin in mature venous thrombi.

Ultrasonography

This technique is commonly used in the assessment of venous thrombosis in the clinical setting and can be used to assess thrombus formation and its resolution in the experimental setting.

High frequency ultrasound (HFUS)

Duplex ultrasonography is a commonly used non-contrast imaging modality for the diagnosis of venous thrombosis in man. Pre-clinical HFUS systems have been developed that allow two-dimensional high-resolution imaging of murine venous thrombi with pixel dimensions in range of 40-70μm ^{48, 49}. IVC thrombus can be identified in transverse and longitudinal planes due to the hyper-echoic nature of the thrombus compared with surrounding blood (Figure 3). The echogenicity of the thrombus periphery is, however, similar to that of blood making accurate segmentation of the thrombus difficult. The absence of flow in the thrombus can be used to improve thrombus segmentation through the use of the colour doppler modality ^{48, 49}. To further improve contrast between the thrombus and blood transpulmonary circulating micro-bubble contrast agents, such as Sonovue can be administered intravenously with the thrombus presenting as a hypo-echoic structure within the vessel lumen⁵⁰.

Figure 3. Visualisation of venous thrombi by high frequency ultrasound.

Transverse view of (A) sham operated mouse with a patent IVC and (B) after IVC ligation with hyperechoic thrombus present in the lumen. Adapted from Aghourain et. al. 2012 ⁴⁸.

Ultrasonic imaging techniques can also be used to provide surrogate measures of thrombus composition. Ultrasound elastography of venous thrombi shows a consistent increase in strain measurements over time indicative of thrombus ‘hardening’, and is taken as thrombus organisation. This technique has been used to accurately estimate clot age in a rat IVC stasis model ⁵¹. Measurements of thrombus size using HFUS have so far been two dimensional limiting data acquisition to either length or cross-sectional area. Available 3D acquisition systems commonly utilised to measure subcutaneous tumour development could be used to reconstruct the thrombus and provide a better measure of thrombus burden ⁵².

Antibody-targeted microbubble contrast agents have been developed to identify fibrin and platelet integrin αIIbβ3 at sites of arterial thrombus formation and these agents may also be applicable to the study of venous thrombosis ^{53, 54}. Alternative agents that provide measures of endothelial cell activation at the sites of atherogenesis through binding cell surface markers such as vascular cell adhesion molecule 1 and intercellular adhesion molecule 1 have also been developed ^{55, 56}. These agents may be of use when studying venous thrombus formation given the role of the endothelium during initiation ¹¹.

Ultrasonic flow probes

Measurements of blood flow in the vessel lumen have been used in the assessment of venous thrombus formation. After induction, blood flow in the thrombosed segment can be monitored using an ultrasonic flow probe over a period of 30-40mins ^{57, 58}. Time to occlusion of the vessel can be used as a measure of thrombus formation. Calculation of time to occlusion may, however, be complicated by residual flow through the thrombosed vessel. Alternatively, flow can be monitored at fixed time-points to observe the degree of vessel stenosis, a technique that has proven robust to the effects of residual blood flow in the arterial system ⁵⁹. Changes in blood flow whilst not providing a direct measure of thrombus size, do act as a measure of vessel stenosis, which may be a more clinically relevant endpoint than measurement of thrombus size. The invasive and terminal nature of this procedure precludes the use of ultrasonic flow measures in longitudinal studies of the same animal.

Intra-vital microscopy

A variety of intra-vital microscopy (IVM) techniques have been used to image experimental thrombus formation in vivo including wide-field video and confocal microscopy. The relatively small depth of field achieved by these techniques has limited the majority of studies to in vivo imaging of thrombosis in the mesenteric and cremaster muscle micro-vasculature. Thrombi, formed in venules can be visualised by fluorescently-labelled antibodies to thrombus constituents such as fibrin and platelets, the accumulation of which allows temporal quantification of thrombus formation ^{60, 61}. Accurate estimation of the spatial resolution of current wide-field IVM systems is complicated by movement and light scatter of tissue during capture, however, the ability of this technique to resolve single platelets suggests that observed resolutions are in the sub-micrometer range⁶². This intra-vital approach has been adapted to study the dynamics of thrombus formation in the femoral vein after localised electrolytic injury of the vessel wall ^{63, 64}. One of the major benefits of IVM based studies of the microvasculature is that multiple thrombotic events can be initiated in a single animal (~10 per mouse), greatly reducing the numbers needed for experimentation. While flow rate and leukocyte rolling differ between the micro and macro-vasculature findings have been largely complementary with respect to thrombosis ^{60, 61}.

IVM has also been used to study the cooperation between neutrophils monocytes and platelets during thrombus initiation in large vessels such as the femoral vein and IVC ³⁷. IVM of the macro-vasculature allows imaging of the luminal vein surface and the thrombus periphery, but because of limitations in tissue penetrance it is not possible to image the thrombus in its entirety. NIR probes have distinct advantages for imaging of macro-vascular thrombosis by IVM owing to the greater penetrance of light through biological tissues in the range of 700-900nm. Confocal IVM with a novel NIR probe has been used to localise fibrin deposition in femoral vein thrombi, demonstrating the potential utility of this technique in informing thrombus structure ³⁹. The terminal nature of the procedure, however, prevents longitudinal measurements in a single animal.

CONCLUSIONS

The most commonly used methods of analysing venous thrombosis in experimental models remain physical measurement of thrombus weight and histological estimates of thrombus size and composition. These techniques are limited to studies that are cross-sectional in design because of their terminal nature. Developments in imaging modalities (such as MRI, microCT and HFUS) have enabled longitudinal in vivo analysis of not only of thrombus burden, but also of thrombus composition that is beginning to rival the information provided by histology. Further developments in these imaging techniques will facilitate concomitant physical, cellular, and molecular analysis of experimental venous thrombi. These will significantly enhance our capacity to investigate mechanisms that regulate the formation and resolution of venous thrombi.

Table 2. Methods of assessing venous thrombus formation and resolution.

	W	H	MRI	CT	RI	HFUS	IVM
Real-time						•	•
Cell tracking							•
Co-localisation		•	•	•	•		•
Large vesselsa	•	•	•	•	•	•	•
Small vesselsb							•
Non-invasive			•	•	•	•
Quantitative	•	•	•	•	•	•	•
Rapid	•			•		•	•
Longitudinal			•	•	•	•
Clinically relevant			•	•	•	•
Reported in mice	•	•	•	•	•	•	•
Resolution (μm)	N/A	N/A	100-500	20-60	350-850	40-70	<1

Abbreviations: W, weight; H, histology; MRI, magnetic resonance imaging; CT, computed tomography; RI, radionucleotide imaging; HFUS, high frequency ultrasound; IVM, intravital microscopy.

^alarge vessels > 1mm in diameter,

^bsmall vessels <100μm in diameter.

SIGNIFICANCE.

Animal models have contributed significantly to our understanding of the mechanisms that govern formation and subsequent resolution of venous thrombi. Here we critically appraise the techniques commonly used to assess thrombosis in these models. Particular attention is paid to imaging modalities that enable longitudinal measurements of thrombus burden and that provide data on composition in a non-invasive manner.

Nonstandard Abbreviations and Acronyms

CT

Computed tomography

DVT

Deep vein thrombosis

FMT

Fluorescence molecular tomography

HFUS

High frequency ultrasound

IVC

Infrarenal vena cava

IVM

Intravital microscopy

MRI

Molecular resonance imaging

NIR

Near infrared

PET

Positron emission tomography

SPECT

Single photon emission computed tomography